Leadership quality trade-off and vote buying: Evidence from the world’s third largest democracy

Last registered on January 31, 2024

View Trial History

Pre-Trial

Trial Information

General Information

Title

Leadership quality trade-off and vote buying: Evidence from the world’s third largest democracy

RCT ID

AEARCTR-0012872

Initial registration date

January 25, 2024

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published

January 31, 2024, 11:41 AM EST

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

Country

Indonesia

Region

Primary Investigator

Name

Naila Shofia

Affiliation

National University of Singapore

Contact Primary Investigator

Other Primary Investigator(s)

PI Name

Piero Stanig

PI Affiliation

Bocconi University

Contact Investigator

Additional Trial Information

Status

On going

Start date

2024-01-09

End date

2024-02-28

Keywords

Behavior, Electoral, Governance

Additional Keywords

JEL code(s)

Secondary IDs

Prior work

This trial does not extend or rely on any prior RCTs.

Abstract

We study the role of the lay conception of democracy in voting behaviour in the context of the 2024 general election in Indonesia. We hypothesise that democracy’s shortcomings might be due to voters inaccurately understanding what the function of democratic elections is, and hence, which candidate qualities they should consider when casting a vote. We carry out a large three-wave panel survey with a representative sample of Indonesian voters with almost four thousand respondents. We randomly assigned information treatment: In the treatment condition, respondents watch an educational animated video explaining the accountability conception of democracy using a metaphor easy to grasp for a layperson. In the control condition, neutral information about general election eligibility and procedures is provided, in a visual format identical to the treatment video. We study a number of outcomes: Preferences regarding leadership qualities, vote buying, and dynasty politics. To measure preferences, we use a conjoint experiment in which respondents choose between fifteen pairs of fully randomised hypothetical legislative candidate profiles. We then estimate how the randomised informational treatment affects the relative weights respondents place on identity markers vis a vis valence and policy positions. Moreover, two list experiments are used to gauge evidence about the prevalence of vote buying during the campaign and in the actual general election. We then estimate whether the informational treatment reduces vote buying and compliance. Finally, we also attempt to understand whether and how much holding an incumbent presidential position leads to an advantage for one’s progeny who are running in a presidential election and gauge whether information treatment changes voter acceptance to dynasty politics.

External Link(s)

Registration Citation

Citation

Shofia, Naila and Piero Stanig. 2024. "Leadership quality trade-off and vote buying: Evidence from the world’s third largest democracy." AEA RCT Registry. January 31. https://doi.org/10.1257/rct.12872-1.0

Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Experimental Details

Interventions

Intervention(s)

Behavioural intervention: We randomly assign informational treatments. In the treatment condition, respondents watch an educational animated video explaining the accountability conception of democracy using a metaphor easy to grasp for a layperson. In one control condition, neutral information about general election eligibility and procedures is provided, in a visual format identical to the treatment video. In a second control condition, respondents do not watch any video.

Measurement intervention: Apart from the behavioral intervention above, we use different methods to experimentally measure the outcomes of interest. First, to measure preferences regarding leadership quality, we are using a conjoint experiment. In this experiment, respondents choose between fifteen pairs of fully randomized hypothetical legislative candidate profiles. Demographic characteristics, identities (religion and ethnicity), valence, and position on divisive issues are randomly assigned to profiles.

Second, we are using list experiment to measure the prevalence of vote buying during the campaign period as well as during the actual election. In list experiment, respondents are asked to state the number of activities they see or experienced from the list we provide them. The experiment randomly vary the number of activities listed for different group. The list given to the control and the treatment group are exactly similar except that the treatment group will also receive a sensitive item in their list (vote buying activities). The magnitude of vote buying is measured by comparing the average number of values stated in the treatment and the control group.

Intervention Start Date

2024-01-25

Intervention End Date

2024-02-07

Primary Outcomes

Primary Outcomes (end points)

Weight assigned to different characteristics of hypothetical candidates
The magnitude of vote buying

Primary Outcomes (explanation)

- Weight assigned to different characteristics of hypothetical candidates : Will be calculated using two methods. First is ACME method (Hainmueller, Hopkins, and Yamamoto, 2014) and second using Random Utility estimation.

-The magnitude of vote buying is estimated by calculating the average value of number elicited by the respondents on different list of activities.

A more detailed elaboration of each method is include in the Pre-Analysis plan document.

Secondary Outcomes

Secondary Outcomes (end points)

Incumbent family advantage
Voter Turnout

Secondary Outcomes (explanation)

See PAP. Incumbent dynastic advantage is measured by vote intention for the challenger ticket that includes the son of the incumbent president as vice-presidential candidate.
Turnout is self-reported in the post-election wave.

Experimental Design

From the pool sample of respondents collected in the baseline (wave 1), the treatment is assigned as follows:

- Wave 2 (Pre-election survey): 2700 respondents will be assigned to the conjoint experiment, while the rest--900 respondents will be assigned to the list experiment. Half of respondents in each experiment will be assigned to the treatment video, while the other half will be assigned to the control video (1800 respondents in total in each treatment)

Wave 3 (Post-election survey): 2700 respondents who were previously assigned to the conjoint experiment will receive the list experiment in this wave. Unlike the pre-election list experiment, this experiment will have three treatment arms: control, treatment 1, and treatment 2. Meanwhile, 900 respondents who were previously assigned to the list experiment will receive the conjoint experiment in this wave. There will be no additional video treatment in this wave, the effect will rely of the treatment the respondents received in the previous wave (pre-election survey)

Experimental Design Details

Randomization Method

Randomization done in office by a computer

Randomization Unit

Individual

Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters

No cluster is applied in this trial

Sample size: planned number of observations

We plan to target 3600 respondents at the end of the third wave survey, but the actual number will depend on several aspect: - The number of responses collected in the baseline - The rate of attrition from one wave to another

Sample size (or number of clusters) by treatment arms

Conjoint experiment : 1800 treatment, 1800 control (of which 900 video controls and 900 no video)
List experiment : 1800 treatment, 1800 control (of which 900 video controls and 900 no video)

Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

Since previous studies in the context and method we are using are limited, we do not have a clear benchmark to calculate the minimum detectable effect size for main outcomes.

Supporting Documents and Materials

IRB